RenderCV Development Context
Gives AI assistants the architecture map, source-code standards, and reference files needed to write and review RenderCV code to the project's own bar.
The skill is pure development-context documentation, using @-references to in-repo source, tests and docs; it requires no network access, credentials, or destructive operations. The only commands (`just format`/`just check`) are local quality checks with minimal risk surface. Deducted for: no explicit data-flow or recovery notes, unverified publisher identity, and inability to statically confirm all referenced file contents are benign.
Internally consistent: specific reference paths (schema/renderer/cli/tests/pyproject/justfile) and standards (typing, docstring order, zero-error `just check`) cross-check against the ruff/ty/pytest configuration in pyproject.toml. Deducted for: no execution verification (static read), no guidance on failure feedback if referenced files are missing or just is unavailable, and undeclared prerequisites (uv/just installation).
Trigger condition is explicit in the description ('Use when writing or reviewing RenderCV code'), with a clear audience (RenderCV contributors) and low false-trigger risk. Deducted for: no declared non-fit boundaries or version applicability, no discussion of Chinese-language support or mainland-China reachability of linked docs (docs.rendercv.com), and no executable outputs.
MIT license is clear; pyproject metadata is complete (version 2.8, changelog/docs URLs, maintainer email); active CI workflows present; content is layered readably (references → guides → standards). Deducted for: the skill itself carries no version/update note, no FAQ or known-limitations section, and maintenance responsibility is only indirectly inferable from the repository.
As a development-context skill it genuinely consolidates code standards, doc paths, and quality-check commands for contributors, though marginal value is modest since experienced contributors could read the repo directly. Deducted for: static review cannot confirm referenced content is complete and current, output is indirect context rather than directly usable deliverable, and comparative-benefit evidence is limited.
The repo supplies auditable primary material: tests directory, coverage badge, CI workflows (issue-triage/PR-review and a test workflow badge), and companion promptfoo evals with a validation grader, corroborating a testing culture; however, these do not directly cover the assessed skill's own key claims (accuracy of referenced content and standards enforcement). Static review executes nothing, so no independent reproduction can be claimed; capped below 5's ceiling.
- This is a purely static source review; no commands or tests were executed, so confidence is low.
- The skill depends on in-repo @-referenced files and the uv/just toolchain; without these prerequisites it may not work as expected, and failure feedback is not documented in the skill.
- The skill itself has no version number or update log; referenced paths or standards may drift as the repository evolves — verify against current code before use.
- Publisher identity is unverified; external docs (docs.rendercv.com etc.) may be unreachable from mainland-China networks.
- The skill applies only to RenderCV codebase development/review; it does not serve end-user resume generation (a different skill's scope).
What does this skill do, and when should you use it?
This is an Agent Skill bundled inside the RenderCV repository at .claude/skills/rendercv-development-context/. It is aimed at contributors, not end users: it pulls codebase architecture docs, testing guidance, JSON Schema notes, and the schema/renderer/cli source directories into the AI's context. The skill body hard-codes RenderCV's source standards — strict Python 3.12+ typing, Google-style docstrings, and a zero-error just check requirement before every commit. Adopt it if you are developing RenderCV itself; it will not help you write a resume with RenderCV.
Loads developer-guide documents via @ references (understanding_rendercv.md, testing.md, _schema.md) plus src/rendercv/schema/, src/rendercv/renderer/, src/rendercv/cli/, tests/, pyproject.toml, and the justfile; provides three how-to guides: add a theme, add a locale, add a social network; enforces coding standards — every function, variable, and class attribute strictly typed, using the type statement, PEP 695 type parameters, pipe unions (str | int), and str | None; mandates Google-style docstrings with optional Why and Example sections ordered as description → Why → Example → Args → Returns → Raises; requires running just format and just check before committing with zero errors from just check, allowing # ty: ignore or #NOQA comments only when no alternative exists.
- A developer contributing a new RenderCV theme who needs the add_theme guide and existing code structure in context
- A contributor adding a new language locale, relying on the add_locale guide and schema context
- A maintainer reviewing pull requests who wants to check typing, docstring format, and just check compliance
- A new maintainer onboarding to the RenderCV codebase who needs to understand the schema/renderer/cli layering quickly
- A developer using an AI coding agent to modify RenderCV source who wants the agent to follow project-specific rules rather than generic Python style
What are this skill's strengths and limitations?
- Standards are concrete and enforceable: strict typing, PEP 695 syntax, and docstring section order are spelled out rather than left vague
- The @ references pull architecture docs and key source directories into context automatically, saving manual retrieval
- Three how-to guides (themes, locales, social networks) cover the most common contribution paths
- Maintained in the official RenderCV repo alongside the code, MIT-licensed
- Only useful for working on RenderCV itself; no value for outside projects
- Depends on in-repo file paths and the just command runner, so it breaks outside this monorepo's context
- The source material does not show whether this skill itself has been evaluated (the promptfoo evals mentioned in the README apply to the user-facing resume skill collection)
How do you install this skill?
The skill ships inside the RenderCV repo as .claude/skills/rendercv-development-context/. The README's install command, npx skills add rendercv/rendercv-skill, installs the user-facing resume skill collection — not this development skill. To use it with Claude Code, clone rendercv/rendercv and keep the repo available so the skill sits under .claude/skills/ (or place it per your client's skills directory convention). The source material gives no more specific standalone install steps.
How do you use this skill?
In a client that supports Agent Skills (e.g. Claude Code), the skill triggers per its description whenever a prompt involves writing or reviewing RenderCV code; you can also invoke it explicitly, e.g. "follow rendercv-development-context standards to add a docstring to this function" or "help me add a new theme per the add_theme guide". It depends on in-repo file references (@docs, @src, @tests), so use it in a workspace rooted at the RenderCV repository.